Line data Source code
1 : /*
2 : - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
3 : SLEPc - Scalable Library for Eigenvalue Problem Computations
4 : Copyright (c) 2002-, Universitat Politecnica de Valencia, Spain
5 :
6 : This file is part of SLEPc.
7 : SLEPc is distributed under a 2-clause BSD license (see LICENSE).
8 : - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
9 : */
10 :
11 : #include <slepc/private/dsimpl.h> /*I "slepcds.h" I*/
12 : #include <slepcblaslapack.h>
13 :
14 : typedef struct {
15 : PetscInt d; /* polynomial degree */
16 : PetscReal *pbc; /* polynomial basis coefficients */
17 : } DS_PEP;
18 :
19 24 : static PetscErrorCode DSAllocate_PEP(DS ds,PetscInt ld)
20 : {
21 24 : DS_PEP *ctx = (DS_PEP*)ds->data;
22 24 : PetscInt i;
23 :
24 24 : PetscFunctionBegin;
25 24 : PetscCheck(ctx->d,PetscObjectComm((PetscObject)ds),PETSC_ERR_ARG_WRONGSTATE,"DSPEP requires specifying the polynomial degree via DSPEPSetDegree()");
26 24 : PetscCall(DSAllocateMat_Private(ds,DS_MAT_X));
27 24 : PetscCall(DSAllocateMat_Private(ds,DS_MAT_Y));
28 109 : for (i=0;i<=ctx->d;i++) PetscCall(DSAllocateMat_Private(ds,DSMatExtra[i]));
29 24 : PetscCall(PetscFree(ds->perm));
30 24 : PetscCall(PetscMalloc1(ld*ctx->d,&ds->perm));
31 24 : PetscFunctionReturn(PETSC_SUCCESS);
32 : }
33 :
34 2 : static PetscErrorCode DSView_PEP(DS ds,PetscViewer viewer)
35 : {
36 2 : DS_PEP *ctx = (DS_PEP*)ds->data;
37 2 : PetscViewerFormat format;
38 2 : PetscInt i;
39 :
40 2 : PetscFunctionBegin;
41 2 : PetscCall(PetscViewerGetFormat(viewer,&format));
42 2 : if (format == PETSC_VIEWER_ASCII_INFO) PetscFunctionReturn(PETSC_SUCCESS);
43 2 : if (format == PETSC_VIEWER_ASCII_INFO_DETAIL) {
44 2 : PetscCall(PetscViewerASCIIPrintf(viewer,"polynomial degree: %" PetscInt_FMT "\n",ctx->d));
45 2 : PetscFunctionReturn(PETSC_SUCCESS);
46 : }
47 0 : for (i=0;i<=ctx->d;i++) PetscCall(DSViewMat(ds,viewer,DSMatExtra[i]));
48 0 : if (ds->state>DS_STATE_INTERMEDIATE) PetscCall(DSViewMat(ds,viewer,DS_MAT_X));
49 0 : PetscFunctionReturn(PETSC_SUCCESS);
50 : }
51 :
52 23 : static PetscErrorCode DSVectors_PEP(DS ds,DSMatType mat,PetscInt *j,PetscReal *rnorm)
53 : {
54 23 : PetscFunctionBegin;
55 23 : PetscCheck(!rnorm,PetscObjectComm((PetscObject)ds),PETSC_ERR_SUP,"Not implemented yet");
56 23 : switch (mat) {
57 : case DS_MAT_X:
58 : break;
59 : case DS_MAT_Y:
60 : break;
61 0 : default:
62 0 : SETERRQ(PetscObjectComm((PetscObject)ds),PETSC_ERR_ARG_OUTOFRANGE,"Invalid mat parameter");
63 : }
64 23 : PetscFunctionReturn(PETSC_SUCCESS);
65 : }
66 :
67 322 : static PetscErrorCode DSSort_PEP(DS ds,PetscScalar *wr,PetscScalar *wi,PetscScalar *rr,PetscScalar *ri,PetscInt *kout)
68 : {
69 322 : DS_PEP *ctx = (DS_PEP*)ds->data;
70 322 : PetscInt n,i,*perm,told;
71 322 : PetscScalar *A;
72 :
73 322 : PetscFunctionBegin;
74 322 : if (!ds->sc) PetscFunctionReturn(PETSC_SUCCESS);
75 322 : n = ds->n*ctx->d;
76 322 : perm = ds->perm;
77 8270 : for (i=0;i<n;i++) perm[i] = i;
78 322 : told = ds->t;
79 322 : ds->t = n; /* force the sorting routines to consider d*n eigenvalues */
80 322 : if (rr) PetscCall(DSSortEigenvalues_Private(ds,rr,ri,perm,PETSC_FALSE));
81 312 : else PetscCall(DSSortEigenvalues_Private(ds,wr,wi,perm,PETSC_FALSE));
82 322 : ds->t = told; /* restore value of t */
83 322 : PetscCall(MatDenseGetArray(ds->omat[DS_MAT_A],&A));
84 8270 : for (i=0;i<n;i++) A[i] = wr[perm[i]];
85 8270 : for (i=0;i<n;i++) wr[i] = A[i];
86 8270 : for (i=0;i<n;i++) A[i] = wi[perm[i]];
87 8270 : for (i=0;i<n;i++) wi[i] = A[i];
88 322 : PetscCall(MatDenseRestoreArray(ds->omat[DS_MAT_A],&A));
89 322 : PetscCall(DSPermuteColumnsTwo_Private(ds,0,n,ds->n,DS_MAT_X,DS_MAT_Y,perm));
90 322 : PetscFunctionReturn(PETSC_SUCCESS);
91 : }
92 :
93 322 : static PetscErrorCode DSSolve_PEP_QZ(DS ds,PetscScalar *wr,PetscScalar *wi)
94 : {
95 322 : DS_PEP *ctx = (DS_PEP*)ds->data;
96 322 : PetscInt i,j,k,off;
97 322 : PetscScalar *A,*B,*W,*X,*U,*Y,*work,*beta,a;
98 322 : const PetscScalar *Ed,*Ei;
99 322 : PetscReal *ca,*cb,*cg,norm,done=1.0;
100 322 : PetscBLASInt info,n,ld,ldd,nd,lwork,one=1,zero=0,cols;
101 322 : PetscBool useggev3=(ds->method==1)?PETSC_TRUE:PETSC_FALSE;
102 :
103 322 : PetscFunctionBegin;
104 322 : PetscCall(PetscBLASIntCast(ds->n*ctx->d,&nd));
105 322 : PetscCall(PetscBLASIntCast(ds->n,&n));
106 322 : PetscCall(PetscBLASIntCast(ds->ld,&ld));
107 322 : PetscCall(PetscBLASIntCast(ds->ld*ctx->d,&ldd));
108 322 : PetscCall(DSAllocateMat_Private(ds,DS_MAT_A));
109 322 : PetscCall(DSAllocateMat_Private(ds,DS_MAT_B));
110 322 : PetscCall(DSAllocateMat_Private(ds,DS_MAT_W));
111 322 : PetscCall(DSAllocateMat_Private(ds,DS_MAT_U));
112 322 : PetscCall(MatDenseGetArray(ds->omat[DS_MAT_A],&A));
113 322 : PetscCall(MatDenseGetArray(ds->omat[DS_MAT_B],&B));
114 :
115 : /* build matrices A and B of the linearization */
116 322 : PetscCall(MatDenseGetArrayRead(ds->omat[DSMatExtra[ctx->d]],&Ed));
117 322 : PetscCall(PetscArrayzero(A,ldd*ldd));
118 322 : if (!ctx->pbc) { /* monomial basis */
119 3274 : for (i=0;i<nd-ds->n;i++) A[i+(i+ds->n)*ldd] = 1.0;
120 848 : for (i=0;i<ctx->d;i++) {
121 566 : PetscCall(MatDenseGetArrayRead(ds->omat[DSMatExtra[i]],&Ei));
122 566 : off = i*ds->n*ldd+(ctx->d-1)*ds->n;
123 6532 : for (j=0;j<ds->n;j++) PetscCall(PetscArraycpy(A+off+j*ldd,Ei+j*ds->ld,ds->n));
124 566 : PetscCall(MatDenseRestoreArrayRead(ds->omat[DSMatExtra[i]],&Ei));
125 : }
126 : } else {
127 40 : ca = ctx->pbc;
128 40 : cb = ca+ctx->d+1;
129 40 : cg = cb+ctx->d+1;
130 746 : for (i=0;i<ds->n;i++) {
131 706 : A[i+(i+ds->n)*ldd] = ca[0];
132 706 : A[i+i*ldd] = cb[0];
133 : }
134 610 : for (;i<nd-ds->n;i++) {
135 570 : j = i/ds->n;
136 570 : A[i+(i+ds->n)*ldd] = ca[j];
137 570 : A[i+i*ldd] = cb[j];
138 570 : A[i+(i-ds->n)*ldd] = cg[j];
139 : }
140 80 : for (i=0;i<ctx->d-2;i++) {
141 40 : PetscCall(MatDenseGetArrayRead(ds->omat[DSMatExtra[i]],&Ei));
142 40 : off = i*ds->n*ldd+(ctx->d-1)*ds->n;
143 610 : for (j=0;j<ds->n;j++)
144 16012 : for (k=0;k<ds->n;k++)
145 15442 : A[off+j*ldd+k] = Ei[j*ds->ld+k]*ca[ctx->d-1];
146 40 : PetscCall(MatDenseRestoreArrayRead(ds->omat[DSMatExtra[i]],&Ei));
147 : }
148 40 : PetscCall(MatDenseGetArrayRead(ds->omat[DSMatExtra[i]],&Ei));
149 40 : off = i*ds->n*ldd+(ctx->d-1)*ds->n;
150 746 : for (j=0;j<ds->n;j++)
151 32222 : for (k=0;k<ds->n;k++)
152 31516 : A[off+j*ldd+k] = Ei[j*ds->ld+k]*ca[ctx->d-1]-Ed[j*ds->ld+k]*cg[ctx->d-1];
153 40 : PetscCall(MatDenseRestoreArrayRead(ds->omat[DSMatExtra[i]],&Ei));
154 40 : i++;
155 40 : PetscCall(MatDenseGetArrayRead(ds->omat[DSMatExtra[i]],&Ei));
156 40 : off = i*ds->n*ldd+(ctx->d-1)*ds->n;
157 746 : for (j=0;j<ds->n;j++)
158 32222 : for (k=0;k<ds->n;k++)
159 31516 : A[off+j*ldd+k] = Ei[j*ds->ld+k]*ca[ctx->d-1]-Ed[j*ds->ld+k]*cb[ctx->d-1];
160 40 : PetscCall(MatDenseRestoreArrayRead(ds->omat[DSMatExtra[i]],&Ei));
161 : }
162 322 : PetscCall(PetscArrayzero(B,ldd*ldd));
163 4590 : for (i=0;i<nd-ds->n;i++) B[i+i*ldd] = 1.0;
164 322 : off = (ctx->d-1)*ds->n*(ldd+1);
165 4002 : for (j=0;j<ds->n;j++) {
166 75000 : for (i=0;i<ds->n;i++) B[off+i+j*ldd] = -Ed[i+j*ds->ld];
167 : }
168 322 : PetscCall(MatDenseRestoreArrayRead(ds->omat[DSMatExtra[ctx->d]],&Ed));
169 :
170 : /* solve generalized eigenproblem */
171 322 : PetscCall(MatDenseGetArray(ds->omat[DS_MAT_W],&W));
172 322 : PetscCall(MatDenseGetArray(ds->omat[DS_MAT_U],&U));
173 322 : lwork = -1;
174 : #if defined(PETSC_USE_COMPLEX)
175 322 : if (useggev3) PetscCallBLAS("LAPACKggev3",LAPACKggev3_("V","V",&nd,A,&ldd,B,&ldd,wr,NULL,U,&ldd,W,&ldd,&a,&lwork,NULL,&info));
176 322 : else PetscCallBLAS("LAPACKggev",LAPACKggev_("V","V",&nd,A,&ldd,B,&ldd,wr,NULL,U,&ldd,W,&ldd,&a,&lwork,NULL,&info));
177 322 : PetscCall(PetscBLASIntCast((PetscInt)PetscRealPart(a),&lwork));
178 322 : PetscCall(DSAllocateWork_Private(ds,lwork+nd,8*nd,0));
179 322 : beta = ds->work;
180 322 : work = ds->work + nd;
181 322 : if (useggev3) PetscCallBLAS("LAPACKggev3",LAPACKggev3_("V","V",&nd,A,&ldd,B,&ldd,wr,beta,U,&ldd,W,&ldd,work,&lwork,ds->rwork,&info));
182 322 : else PetscCallBLAS("LAPACKggev",LAPACKggev_("V","V",&nd,A,&ldd,B,&ldd,wr,beta,U,&ldd,W,&ldd,work,&lwork,ds->rwork,&info));
183 : #else
184 : if (useggev3) PetscCallBLAS("LAPACKggev3",LAPACKggev3_("V","V",&nd,A,&ldd,B,&ldd,wr,wi,NULL,U,&ldd,W,&ldd,&a,&lwork,&info));
185 : else PetscCallBLAS("LAPACKggev",LAPACKggev_("V","V",&nd,A,&ldd,B,&ldd,wr,wi,NULL,U,&ldd,W,&ldd,&a,&lwork,&info));
186 : PetscCall(PetscBLASIntCast((PetscInt)a,&lwork));
187 : PetscCall(DSAllocateWork_Private(ds,lwork+nd,0,0));
188 : beta = ds->work;
189 : work = ds->work + nd;
190 : if (useggev3) PetscCallBLAS("LAPACKggev3",LAPACKggev3_("V","V",&nd,A,&ldd,B,&ldd,wr,wi,beta,U,&ldd,W,&ldd,work,&lwork,&info));
191 : else PetscCallBLAS("LAPACKggev",LAPACKggev_("V","V",&nd,A,&ldd,B,&ldd,wr,wi,beta,U,&ldd,W,&ldd,work,&lwork,&info));
192 : #endif
193 322 : SlepcCheckLapackInfo(useggev3?"ggev3":"ggev",info);
194 322 : PetscCall(MatDenseRestoreArray(ds->omat[DS_MAT_A],&A));
195 322 : PetscCall(MatDenseRestoreArray(ds->omat[DS_MAT_B],&B));
196 :
197 : /* copy eigenvalues */
198 8270 : for (i=0;i<nd;i++) {
199 7948 : if (beta[i]==0.0) wr[i] = (PetscRealPart(wr[i])>0.0)? PETSC_MAX_REAL: PETSC_MIN_REAL;
200 7947 : else wr[i] /= beta[i];
201 : #if !defined(PETSC_USE_COMPLEX)
202 : if (beta[i]==0.0) wi[i] = 0.0;
203 : else wi[i] /= beta[i];
204 : #else
205 7948 : if (wi) wi[i] = 0.0;
206 : #endif
207 : }
208 :
209 : /* copy and normalize eigenvectors */
210 322 : PetscCall(MatDenseGetArray(ds->omat[DS_MAT_X],&X));
211 322 : PetscCall(MatDenseGetArray(ds->omat[DS_MAT_Y],&Y));
212 8270 : for (j=0;j<nd;j++) {
213 7948 : PetscCall(PetscArraycpy(X+j*ds->ld,W+j*ldd,ds->n));
214 7948 : PetscCall(PetscArraycpy(Y+j*ds->ld,U+ds->n*(ctx->d-1)+j*ldd,ds->n));
215 : }
216 322 : PetscCall(MatDenseRestoreArray(ds->omat[DS_MAT_W],&W));
217 322 : PetscCall(MatDenseRestoreArray(ds->omat[DS_MAT_U],&U));
218 8270 : for (j=0;j<nd;j++) {
219 7948 : cols = 1;
220 7948 : norm = BLASnrm2_(&n,X+j*ds->ld,&one);
221 : #if !defined(PETSC_USE_COMPLEX)
222 : if (wi[j] != 0.0) {
223 : norm = SlepcAbsEigenvalue(norm,BLASnrm2_(&n,X+(j+1)*ds->ld,&one));
224 : cols = 2;
225 : }
226 : #endif
227 7948 : PetscCallBLAS("LAPACKlascl",LAPACKlascl_("G",&zero,&zero,&norm,&done,&n,&cols,X+j*ds->ld,&ld,&info));
228 7948 : SlepcCheckLapackInfo("lascl",info);
229 7948 : norm = BLASnrm2_(&n,Y+j*ds->ld,&one);
230 : #if !defined(PETSC_USE_COMPLEX)
231 : if (wi[j] != 0.0) norm = SlepcAbsEigenvalue(norm,BLASnrm2_(&n,Y+(j+1)*ds->ld,&one));
232 : #endif
233 7948 : PetscCallBLAS("LAPACKlascl",LAPACKlascl_("G",&zero,&zero,&norm,&done,&n,&cols,Y+j*ds->ld,&ld,&info));
234 7948 : SlepcCheckLapackInfo("lascl",info);
235 : #if !defined(PETSC_USE_COMPLEX)
236 : if (wi[j] != 0.0) j++;
237 : #endif
238 : }
239 322 : PetscCall(MatDenseRestoreArray(ds->omat[DS_MAT_X],&X));
240 322 : PetscCall(MatDenseRestoreArray(ds->omat[DS_MAT_Y],&Y));
241 322 : PetscFunctionReturn(PETSC_SUCCESS);
242 : }
243 :
244 : #if !defined(PETSC_HAVE_MPIUNI)
245 52 : static PetscErrorCode DSSynchronize_PEP(DS ds,PetscScalar eigr[],PetscScalar eigi[])
246 : {
247 52 : DS_PEP *ctx = (DS_PEP*)ds->data;
248 52 : PetscInt ld=ds->ld,k=0;
249 52 : PetscMPIInt ldnd,rank,off=0,size,dn;
250 52 : PetscScalar *X,*Y;
251 :
252 52 : PetscFunctionBegin;
253 52 : if (ds->state>=DS_STATE_CONDENSED) k += 2*ctx->d*ds->n*ld;
254 52 : if (eigr) k += ctx->d*ds->n;
255 52 : if (eigi) k += ctx->d*ds->n;
256 52 : PetscCall(DSAllocateWork_Private(ds,k,0,0));
257 52 : PetscCall(PetscMPIIntCast(k*sizeof(PetscScalar),&size));
258 52 : PetscCall(PetscMPIIntCast(ds->n*ctx->d*ld,&ldnd));
259 52 : PetscCall(PetscMPIIntCast(ctx->d*ds->n,&dn));
260 52 : if (ds->state>=DS_STATE_CONDENSED) {
261 52 : PetscCall(MatDenseGetArray(ds->omat[DS_MAT_X],&X));
262 52 : PetscCall(MatDenseGetArray(ds->omat[DS_MAT_Y],&Y));
263 : }
264 52 : PetscCallMPI(MPI_Comm_rank(PetscObjectComm((PetscObject)ds),&rank));
265 52 : if (!rank) {
266 26 : if (ds->state>=DS_STATE_CONDENSED) {
267 26 : PetscCallMPI(MPI_Pack(X,ldnd,MPIU_SCALAR,ds->work,size,&off,PetscObjectComm((PetscObject)ds)));
268 26 : PetscCallMPI(MPI_Pack(Y,ldnd,MPIU_SCALAR,ds->work,size,&off,PetscObjectComm((PetscObject)ds)));
269 : }
270 26 : if (eigr) PetscCallMPI(MPI_Pack(eigr,dn,MPIU_SCALAR,ds->work,size,&off,PetscObjectComm((PetscObject)ds)));
271 : #if !defined(PETSC_USE_COMPLEX)
272 : if (eigi) PetscCallMPI(MPI_Pack(eigi,dn,MPIU_SCALAR,ds->work,size,&off,PetscObjectComm((PetscObject)ds)));
273 : #endif
274 : }
275 104 : PetscCallMPI(MPI_Bcast(ds->work,size,MPI_BYTE,0,PetscObjectComm((PetscObject)ds)));
276 52 : if (rank) {
277 26 : if (ds->state>=DS_STATE_CONDENSED) {
278 26 : PetscCallMPI(MPI_Unpack(ds->work,size,&off,X,ldnd,MPIU_SCALAR,PetscObjectComm((PetscObject)ds)));
279 26 : PetscCallMPI(MPI_Unpack(ds->work,size,&off,Y,ldnd,MPIU_SCALAR,PetscObjectComm((PetscObject)ds)));
280 : }
281 26 : if (eigr) PetscCallMPI(MPI_Unpack(ds->work,size,&off,eigr,dn,MPIU_SCALAR,PetscObjectComm((PetscObject)ds)));
282 : #if !defined(PETSC_USE_COMPLEX)
283 : if (eigi) PetscCallMPI(MPI_Unpack(ds->work,size,&off,eigi,dn,MPIU_SCALAR,PetscObjectComm((PetscObject)ds)));
284 : #endif
285 : }
286 52 : if (ds->state>=DS_STATE_CONDENSED) {
287 52 : PetscCall(MatDenseRestoreArray(ds->omat[DS_MAT_X],&X));
288 52 : PetscCall(MatDenseRestoreArray(ds->omat[DS_MAT_Y],&Y));
289 : }
290 52 : PetscFunctionReturn(PETSC_SUCCESS);
291 : }
292 : #endif
293 :
294 25 : static PetscErrorCode DSPEPSetDegree_PEP(DS ds,PetscInt d)
295 : {
296 25 : DS_PEP *ctx = (DS_PEP*)ds->data;
297 :
298 25 : PetscFunctionBegin;
299 25 : PetscCheck(d>=0,PetscObjectComm((PetscObject)ds),PETSC_ERR_ARG_OUTOFRANGE,"The degree must be a non-negative integer");
300 25 : PetscCheck(d<DS_NUM_EXTRA,PetscObjectComm((PetscObject)ds),PETSC_ERR_ARG_OUTOFRANGE,"Only implemented for polynomials of degree at most %d",DS_NUM_EXTRA-1);
301 25 : ctx->d = d;
302 25 : PetscFunctionReturn(PETSC_SUCCESS);
303 : }
304 :
305 : /*@
306 : DSPEPSetDegree - Sets the polynomial degree for a DSPEP.
307 :
308 : Logically Collective
309 :
310 : Input Parameters:
311 : + ds - the direct solver context
312 : - d - the degree
313 :
314 : Level: intermediate
315 :
316 : .seealso: DSPEPGetDegree()
317 : @*/
318 25 : PetscErrorCode DSPEPSetDegree(DS ds,PetscInt d)
319 : {
320 25 : PetscFunctionBegin;
321 25 : PetscValidHeaderSpecific(ds,DS_CLASSID,1);
322 100 : PetscValidLogicalCollectiveInt(ds,d,2);
323 25 : PetscTryMethod(ds,"DSPEPSetDegree_C",(DS,PetscInt),(ds,d));
324 25 : PetscFunctionReturn(PETSC_SUCCESS);
325 : }
326 :
327 1337 : static PetscErrorCode DSPEPGetDegree_PEP(DS ds,PetscInt *d)
328 : {
329 1337 : DS_PEP *ctx = (DS_PEP*)ds->data;
330 :
331 1337 : PetscFunctionBegin;
332 1337 : *d = ctx->d;
333 1337 : PetscFunctionReturn(PETSC_SUCCESS);
334 : }
335 :
336 : /*@
337 : DSPEPGetDegree - Returns the polynomial degree for a DSPEP.
338 :
339 : Not Collective
340 :
341 : Input Parameter:
342 : . ds - the direct solver context
343 :
344 : Output Parameters:
345 : . d - the degree
346 :
347 : Level: intermediate
348 :
349 : .seealso: DSPEPSetDegree()
350 : @*/
351 1337 : PetscErrorCode DSPEPGetDegree(DS ds,PetscInt *d)
352 : {
353 1337 : PetscFunctionBegin;
354 1337 : PetscValidHeaderSpecific(ds,DS_CLASSID,1);
355 1337 : PetscAssertPointer(d,2);
356 1337 : PetscUseMethod(ds,"DSPEPGetDegree_C",(DS,PetscInt*),(ds,d));
357 1337 : PetscFunctionReturn(PETSC_SUCCESS);
358 : }
359 :
360 12 : static PetscErrorCode DSPEPSetCoefficients_PEP(DS ds,PetscReal *pbc)
361 : {
362 12 : DS_PEP *ctx = (DS_PEP*)ds->data;
363 12 : PetscInt i;
364 :
365 12 : PetscFunctionBegin;
366 12 : PetscCheck(ctx->d,PetscObjectComm((PetscObject)ds),PETSC_ERR_ARG_WRONGSTATE,"Must first specify the polynomial degree via DSPEPSetDegree()");
367 12 : if (ctx->pbc) PetscCall(PetscFree(ctx->pbc));
368 12 : PetscCall(PetscMalloc1(3*(ctx->d+1),&ctx->pbc));
369 156 : for (i=0;i<3*(ctx->d+1);i++) ctx->pbc[i] = pbc[i];
370 12 : ds->state = DS_STATE_RAW;
371 12 : PetscFunctionReturn(PETSC_SUCCESS);
372 : }
373 :
374 : /*@C
375 : DSPEPSetCoefficients - Sets the polynomial basis coefficients for a DSPEP.
376 :
377 : Logically Collective
378 :
379 : Input Parameters:
380 : + ds - the direct solver context
381 : - pbc - the polynomial basis coefficients
382 :
383 : Notes:
384 : This function is required only in the case of a polynomial specified in a
385 : non-monomial basis, to provide the coefficients that will be used
386 : during the linearization, multiplying the identity blocks on the three main
387 : diagonal blocks. Depending on the polynomial basis (Chebyshev, Legendre, ...)
388 : the coefficients must be different.
389 :
390 : There must be a total of 3*(d+1) coefficients, where d is the degree of the
391 : polynomial. The coefficients are arranged in three groups, alpha, beta, and
392 : gamma, according to the definition of the three-term recurrence. In the case
393 : of the monomial basis, alpha=1 and beta=gamma=0, in which case it is not
394 : necessary to invoke this function.
395 :
396 : Level: advanced
397 :
398 : .seealso: DSPEPGetCoefficients(), DSPEPSetDegree()
399 : @*/
400 12 : PetscErrorCode DSPEPSetCoefficients(DS ds,PetscReal *pbc)
401 : {
402 12 : PetscFunctionBegin;
403 12 : PetscValidHeaderSpecific(ds,DS_CLASSID,1);
404 12 : PetscTryMethod(ds,"DSPEPSetCoefficients_C",(DS,PetscReal*),(ds,pbc));
405 12 : PetscFunctionReturn(PETSC_SUCCESS);
406 : }
407 :
408 1 : static PetscErrorCode DSPEPGetCoefficients_PEP(DS ds,PetscReal **pbc)
409 : {
410 1 : DS_PEP *ctx = (DS_PEP*)ds->data;
411 1 : PetscInt i;
412 :
413 1 : PetscFunctionBegin;
414 1 : PetscCheck(ctx->d,PetscObjectComm((PetscObject)ds),PETSC_ERR_ARG_WRONGSTATE,"Must first specify the polynomial degree via DSPEPSetDegree()");
415 1 : PetscCall(PetscCalloc1(3*(ctx->d+1),pbc));
416 1 : if (ctx->pbc) for (i=0;i<3*(ctx->d+1);i++) (*pbc)[i] = ctx->pbc[i];
417 4 : else for (i=0;i<ctx->d+1;i++) (*pbc)[i] = 1.0;
418 1 : PetscFunctionReturn(PETSC_SUCCESS);
419 : }
420 :
421 : /*@C
422 : DSPEPGetCoefficients - Returns the polynomial basis coefficients for a DSPEP.
423 :
424 : Not Collective
425 :
426 : Input Parameter:
427 : . ds - the direct solver context
428 :
429 : Output Parameters:
430 : . pbc - the polynomial basis coefficients
431 :
432 : Note:
433 : The returned array has length 3*(d+1) and should be freed by the user.
434 :
435 : Fortran Notes:
436 : The calling sequence from Fortran is
437 : .vb
438 : DSPEPGetCoefficients(eps,pbc,ierr)
439 : double precision pbc(d+1) output
440 : .ve
441 :
442 : Level: advanced
443 :
444 : .seealso: DSPEPSetCoefficients()
445 : @*/
446 1 : PetscErrorCode DSPEPGetCoefficients(DS ds,PetscReal **pbc)
447 : {
448 1 : PetscFunctionBegin;
449 1 : PetscValidHeaderSpecific(ds,DS_CLASSID,1);
450 1 : PetscAssertPointer(pbc,2);
451 1 : PetscUseMethod(ds,"DSPEPGetCoefficients_C",(DS,PetscReal**),(ds,pbc));
452 1 : PetscFunctionReturn(PETSC_SUCCESS);
453 : }
454 :
455 24 : static PetscErrorCode DSDestroy_PEP(DS ds)
456 : {
457 24 : DS_PEP *ctx = (DS_PEP*)ds->data;
458 :
459 24 : PetscFunctionBegin;
460 24 : if (ctx->pbc) PetscCall(PetscFree(ctx->pbc));
461 24 : PetscCall(PetscFree(ds->data));
462 24 : PetscCall(PetscObjectComposeFunction((PetscObject)ds,"DSPEPSetDegree_C",NULL));
463 24 : PetscCall(PetscObjectComposeFunction((PetscObject)ds,"DSPEPGetDegree_C",NULL));
464 24 : PetscCall(PetscObjectComposeFunction((PetscObject)ds,"DSPEPSetCoefficients_C",NULL));
465 24 : PetscCall(PetscObjectComposeFunction((PetscObject)ds,"DSPEPGetCoefficients_C",NULL));
466 24 : PetscFunctionReturn(PETSC_SUCCESS);
467 : }
468 :
469 1050 : static PetscErrorCode DSMatGetSize_PEP(DS ds,DSMatType t,PetscInt *rows,PetscInt *cols)
470 : {
471 1050 : DS_PEP *ctx = (DS_PEP*)ds->data;
472 :
473 1050 : PetscFunctionBegin;
474 1050 : PetscCheck(ctx->d,PetscObjectComm((PetscObject)ds),PETSC_ERR_ARG_WRONGSTATE,"DSPEP requires specifying the polynomial degree via DSPEPSetDegree()");
475 1050 : *rows = ds->n;
476 1050 : if (t==DS_MAT_A || t==DS_MAT_B || t==DS_MAT_W || t==DS_MAT_U) *rows *= ctx->d;
477 1050 : *cols = ds->n;
478 1050 : if (t==DS_MAT_A || t==DS_MAT_B || t==DS_MAT_W || t==DS_MAT_U || t==DS_MAT_X || t==DS_MAT_Y) *cols *= ctx->d;
479 1050 : PetscFunctionReturn(PETSC_SUCCESS);
480 : }
481 :
482 : /*MC
483 : DSPEP - Dense Polynomial Eigenvalue Problem.
484 :
485 : Level: beginner
486 :
487 : Notes:
488 : The problem is expressed as P(lambda)*x = 0, where P(.) is a matrix
489 : polynomial of degree d. The eigenvalues lambda are the arguments
490 : returned by DSSolve().
491 :
492 : The degree of the polynomial, d, can be set with DSPEPSetDegree(), with
493 : the first d+1 extra matrices of the DS defining the matrix polynomial. By
494 : default, the polynomial is expressed in the monomial basis, but a
495 : different basis can be used by setting the corresponding coefficients
496 : via DSPEPSetCoefficients().
497 :
498 : The problem is solved via linearization, by building a pencil (A,B) of
499 : size p*n and solving the corresponding GNHEP.
500 :
501 : Used DS matrices:
502 : + DS_MAT_Ex - coefficients of the matrix polynomial
503 : . DS_MAT_A - (workspace) first matrix of the linearization
504 : . DS_MAT_B - (workspace) second matrix of the linearization
505 : . DS_MAT_W - (workspace) right eigenvectors of the linearization
506 : - DS_MAT_U - (workspace) left eigenvectors of the linearization
507 :
508 : Implemented methods:
509 : . 0 - QZ iteration on the linearization (_ggev)
510 :
511 : .seealso: DSCreate(), DSSetType(), DSType, DSPEPSetDegree(), DSPEPSetCoefficients()
512 : M*/
513 24 : SLEPC_EXTERN PetscErrorCode DSCreate_PEP(DS ds)
514 : {
515 24 : DS_PEP *ctx;
516 :
517 24 : PetscFunctionBegin;
518 24 : PetscCall(PetscNew(&ctx));
519 24 : ds->data = (void*)ctx;
520 :
521 24 : ds->ops->allocate = DSAllocate_PEP;
522 24 : ds->ops->view = DSView_PEP;
523 24 : ds->ops->vectors = DSVectors_PEP;
524 24 : ds->ops->solve[0] = DSSolve_PEP_QZ;
525 : #if !defined(SLEPC_MISSING_LAPACK_GGES3)
526 24 : ds->ops->solve[1] = DSSolve_PEP_QZ;
527 : #endif
528 24 : ds->ops->sort = DSSort_PEP;
529 : #if !defined(PETSC_HAVE_MPIUNI)
530 24 : ds->ops->synchronize = DSSynchronize_PEP;
531 : #endif
532 24 : ds->ops->destroy = DSDestroy_PEP;
533 24 : ds->ops->matgetsize = DSMatGetSize_PEP;
534 24 : PetscCall(PetscObjectComposeFunction((PetscObject)ds,"DSPEPSetDegree_C",DSPEPSetDegree_PEP));
535 24 : PetscCall(PetscObjectComposeFunction((PetscObject)ds,"DSPEPGetDegree_C",DSPEPGetDegree_PEP));
536 24 : PetscCall(PetscObjectComposeFunction((PetscObject)ds,"DSPEPSetCoefficients_C",DSPEPSetCoefficients_PEP));
537 24 : PetscCall(PetscObjectComposeFunction((PetscObject)ds,"DSPEPGetCoefficients_C",DSPEPGetCoefficients_PEP));
538 24 : PetscFunctionReturn(PETSC_SUCCESS);
539 : }
|